Data

Imports

import os
import pandas as pd
import numpy as np

Merging data from 3 csvs

import pandas as pd

# Load the CSV files
df_ss = pd.read_csv('spreadspoke_scores.csv')
df_teams = pd.read_csv('nfl_teams.csv')
df_stads = pd.read_csv('nfl_stadiums.csv', encoding='ISO-8859-1')

df_teams = df_teams.rename(columns={'team_name':'team_home'})
merged_df = pd.merge(df_teams, df_ss, on='team_home', how='inner')# Print the column names of both DataFrames

df_stads = df_stads.rename(columns={'stadium_name': 'stadium'})
df = pd.merge(merged_df, df_stads, on='stadium', how='inner')

## Schedule date transformed to date type, sorted df by schedule date
df['schedule_date'] = pd.to_datetime(df['schedule_date'], format='%m/%d/%Y', errors='coerce')
df = df.sort_values(by='schedule_date').reset_index(drop=True)

## This value should be numeric but is object type
df['over_under_line'] = pd.to_numeric(df['over_under_line'], errors='coerce')


df.tail()

	team_home	team_name_short	team_id	team_id_pfr	team_conference	team_division	team_conference_pre2002	team_division_pre2002	schedule_date	schedule_season	...	stadium_weather_station_zipcode	stadium_weather_type	stadium_capacity	stadium_surface	stadium_weather_station	stadium_weather_station_name	stadium_latitude	stadium_longitude	stadium_azimuthangle	stadium_elevation
14067	Indianapolis Colts	Colts	IND	CLT	AFC	AFC South	AFC	AFC East	2025-01-05	2024	...	46225	indoor	63,000	FieldTurf	US1INMR0076	INDIANAPOLIS 6.8 NNE, IN US	39.760056	-86.162806	26.6	227.1
14068	New York Jets	Jets	NYJ	NYJ	NFC	AFC East	NFC	AFC East	2025-01-05	2024	...	7073	cold	82,500	FieldTurf	USW00014734	NEWARK LIBERTY INTERNATIONAL AIRPORT, NJ US	40.813528	-74.074361	345.5	2.1
14069	Baltimore Ravens	Ravens	BAL	RAV	AFC	AFC North	AFC	AFC Central	2025-01-05	2024	...	21230	moderate	71,008	FieldTurf	USW00093721	BALTIMORE WASHINGTON INTERNATIONAL AIRPORT, MD US	39.278056	-76.622778	289.5	47.5
14070	Green Bay Packers	Packers	GB	GNB	NFC	NFC North	NFC	NFC Central	2025-01-05	2024	...	54304	cold	80,735	Grass	USW00014898	GREEN BAY AUSTIN STRAUBEL INTERNATIONAL AIRPOR...	44.501389	-88.062222	0.0	209.4
14071	Tennessee Titans	Titans	TEN	OTI	AFC	AFC South	AFC	AFC Central	2025-01-05	2024	...	37213	moderate	69,143	Grass	USW00013897	NASHVILLE INTERNATIONAL AIRPORT, TN US	36.166389	-86.771389	334.6	182.9

5 rows × 39 columns

Object Data

df_cat = df.select_dtypes(include=['object', 'category'])
df_cat = df_cat.fillna('Unknown')
df_cat.drop(columns=['team_name_short', 'team_id', 'team_id_pfr','stadium_address','stadium_weather_station_zipcode','stadium_weather_station_name','stadium_weather_station'], inplace=True)  ### IRRELEVANT
df_cat.dtypes

team_home                  object
team_conference            object
team_division              object
team_conference_pre2002    object
team_division_pre2002      object
schedule_week              object
team_away                  object
team_favorite_id           object
stadium                    object
weather_detail             object
stadium_location           object
stadium_type               object
stadium_weather_type       object
stadium_capacity           object
stadium_surface            object
dtype: object

Float Data

df_num = df.select_dtypes(include=['float64'])
df_num[df_num.columns] = df_num.fillna(df_num.mean())
df_num.head()

	score_home	score_away	spread_favorite	over_under_line	weather_temperature	weather_wind_mph	weather_humidity	stadium_open	stadium_close	stadium_latitude	stadium_longitude	stadium_azimuthangle	stadium_elevation
0	14.0	23.0	-5.372946	42.282329	83.0	6.0	71.0	1946.000000	1986.000000	25.790500	-80.316300	192.485631	8.800000
1	45.0	7.0	-5.372946	42.282329	81.0	7.0	70.0	1978.466085	2002.689877	38.021551	-89.763382	192.485631	187.972944
2	27.0	7.0	-5.372946	42.282329	70.0	7.0	82.0	1978.466085	2002.689877	38.021551	-89.763382	192.485631	187.972944
3	14.0	19.0	-5.372946	42.282329	82.0	11.0	78.0	1946.000000	1986.000000	25.790500	-80.316300	192.485631	8.800000
4	31.0	0.0	-5.372946	42.282329	77.0	6.0	82.0	1978.466085	2002.689877	38.021551	-89.763382	192.485631	187.972944

Boolean Data

df_bool = df.select_dtypes(include=[bool]).astype(int)
df_bool

	schedule_playoff	stadium_neutral
0	0	0
1	0	0
2	0	0
3	0	0
4	0	0
...	...	...
14067	0	0
14068	0	0
14069	0	0
14070	0	0
14071	0	0

14072 rows × 2 columns

Integer Data

df_int = df.select_dtypes(include=['int64'])
df_int

	schedule_season
0	1966
1	1966
2	1966
3	1966
4	1966
...	...
14067	2024
14068	2024
14069	2024
14070	2024
14071	2024

14072 rows × 1 columns

cleaned Data

nfl1 = pd.concat([df_cat, df_num], axis=1)
nfl2 = pd.concat([ df_int, df_bool],axis=1)
nfl = pd.concat([nfl1,nfl2],axis=1)
nfl['score_home'].isna().sum()

0

Visualized Data

Violin Plot

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns


df_cat['score_home'] = nfl['score_home']
df_cat['score_away'] = nfl['score_away']

for feature in df_cat.columns:
    if feature != 'score_home' and feature != 'score_away':
        plt.figure(figsize=(12, 12))
        
        ax1 = plt.subplot(2, 1, 1)
        sns.violinplot(data=df_cat, x=feature, y='score_home', ax=ax1, color='red')
        ax1.set_title(f'home score and away score vs {feature}', fontsize=16)
        ax1.set_xlabel('')
        ax1.set_ylabel('home score', fontsize=14)
        ax1.set_xticklabels([])  # Remove x-axis tick labels for score_home plot
        ax1.grid(True)

        ax2 = plt.subplot(2, 1, 2)
        sns.violinplot(data=df_cat, x=feature, y='score_away', ax=ax2, color='blue')
        ax2.set_xlabel(feature, fontsize=14)
        ax2.set_ylabel('away score', fontsize=14)
        ax2.tick_params(axis='x', rotation=45)
        ax2.grid(True)
        
        plt.tight_layout()
        plt.show()

df_num.dtypes

score_home              float64
score_away              float64
spread_favorite         float64
over_under_line         float64
weather_temperature     float64
weather_wind_mph        float64
weather_humidity        float64
stadium_open            float64
stadium_close           float64
stadium_latitude        float64
stadium_longitude       float64
stadium_azimuthangle    float64
stadium_elevation       float64
dtype: object

import seaborn as sns
import matplotlib.pyplot as plt

# Select columns for the pair plot
columns_to_plot = ['score_home', 'score_away', 'over_under_line', 'spread_favorite', 'weather_temperature', 'weather_wind_mph', 'weather_humidity']

# Create a new DataFrame with only the selected columns
plot_df = nfl[columns_to_plot]

# Create the pair plot
sns.pairplot(plot_df)
plt.suptitle('Pair Plot of Selected Columns', y=1.02)
plt.show()

Randon Forest Model

import pandas as pd
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_squared_error
import time

# Measure runtime
start_time = time.time()

# Separate the last 271 rows for prediction
nfl_train = nfl.iloc[:-271]  
nfl_predict = nfl.iloc[-271:]  

# Convert categorical columns to numerical using one-hot encoding
nfl_train_encoded = pd.get_dummies(nfl_train)
nfl_predict_encoded = pd.get_dummies(nfl_predict)

# Ensure both training and prediction data have the same columns
nfl_predict_encoded = nfl_predict_encoded.reindex(columns=nfl_train_encoded.columns, fill_value=0)

X_train = nfl_train_encoded.drop(columns=['score_home', 'score_away'])  
y_train_home = nfl_train_encoded['score_home']  
y_train_away = nfl_train_encoded['score_away']  

X_predict = nfl_predict_encoded.drop(columns=['score_home', 'score_away'])  

rf_model_home = RandomForestRegressor(n_estimators=100, random_state=42)
rf_model_away = RandomForestRegressor(n_estimators=100, random_state=42)
rf_model_home.fit(X_train, y_train_home)
rf_model_away.fit(X_train, y_train_away)
predicted_home_scores = rf_model_home.predict(X_predict)
predicted_away_scores = rf_model_away.predict(X_predict)
nfl_predict['predicted_score_home'] = predicted_home_scores
nfl_predict['predicted_score_away'] = predicted_away_scores

# Calculate Mean Squared Error
mse_home = mean_squared_error(nfl_predict['score_home'], predicted_home_scores)
mse_away = mean_squared_error(nfl_predict['score_away'], predicted_away_scores)

# Measure runtime
end_time = time.time()
runtime = end_time - start_time

print(nfl_predict[['predicted_score_home', 'predicted_score_away']])
print(f"Mean Squared Error (Home Scores): {mse_home}")
print(f"Mean Squared Error (Away Scores): {mse_away}")
print(f"Runtime: {runtime:.2f} seconds")


#######  OUTPUT  #######
#   predicted_score_home  predicted_score_away
#   13801                 17.57                 20.50
#   13802                 19.12                 18.85
#   13803                 19.31                 19.86
#   13804                 23.37                 18.59
#   13805                 21.11                 16.22
#   ...                     ...                   ...
#   14067                 23.76                 19.86
#   14068                 16.41                 21.15
#   14069                 21.40                 21.36
#   14070                 19.96                 18.44
#   14071                 21.88                 18.90
#   
#   
#   
#   [271 rows x 2 columns]
#   Mean Squared Error (Home Scores): 6.841222137206541
#   Mean Squared Error (Away Scores): 2.8940572362235955
#   Runtime: 76.65 seconds

       predicted_score_home  predicted_score_away
13801                 17.57                 20.50
13802                 19.12                 18.85
13803                 19.31                 19.86
13804                 23.37                 18.59
13805                 21.11                 16.22
...                     ...                   ...
14067                 23.76                 19.86
14068                 16.41                 21.15
14069                 21.40                 21.36
14070                 19.96                 18.44
14071                 21.88                 18.90

[271 rows x 2 columns]
Mean Squared Error (Home Scores): 6.841222137206541
Mean Squared Error (Away Scores): 2.8940572362235955
Runtime: 77.36 seconds

/var/folders/xc/yrj64zn13bjfv7vdfm67_kb40000gn/T/ipykernel_10625/4112753878.py:32: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  nfl_predict['predicted_score_home'] = predicted_home_scores
/var/folders/xc/yrj64zn13bjfv7vdfm67_kb40000gn/T/ipykernel_10625/4112753878.py:33: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  nfl_predict['predicted_score_away'] = predicted_away_scores

Predicted Scores

original_info = df.tail(271)[['schedule_date', 'team_home', 'team_away']].reset_index(drop=True)

# Combine the original info with predictions
result_df = pd.concat([original_info, nfl_predict[['predicted_score_home', 'predicted_score_away']].reset_index(drop=True)], axis=1)

result_df

	schedule_date	team_home	team_away	predicted_score_home	predicted_score_away
0	2024-09-05	Kansas City Chiefs	Baltimore Ravens	17.57	20.50
1	2024-09-08	Tampa Bay Buccaneers	Washington Commanders	19.12	18.85
2	2024-09-08	Cincinnati Bengals	New England Patriots	19.31	19.86
3	2024-09-08	Buffalo Bills	Arizona Cardinals	23.37	18.59
4	2024-09-08	Los Angeles Chargers	Las Vegas Raiders	21.11	16.22
...	...	...	...	...	...
266	2025-01-05	Indianapolis Colts	Jacksonville Jaguars	23.76	19.86
267	2025-01-05	New York Jets	Miami Dolphins	16.41	21.15
268	2025-01-05	Baltimore Ravens	Cleveland Browns	21.40	21.36
269	2025-01-05	Green Bay Packers	Chicago Bears	19.96	18.44
270	2025-01-05	Tennessee Titans	Houston Texans	21.88	18.90

271 rows × 5 columns

Plot of Predicted Scores

plt.figure(figsize=(15, 8))
plt.scatter(result_df['schedule_date'], result_df['predicted_score_home'], label='Predicted Score Home', color='blue', marker='o', alpha=0.7)
plt.scatter(result_df['schedule_date'], result_df['predicted_score_away'], label='Predicted Score Away', color='green', marker='x', alpha=0.7)

plt.title('Predicted Scores vs. Schedule Date')
plt.xlabel('Schedule Date')
plt.ylabel('Predicted Score')
plt.legend()
plt.grid(True)

plt.show()

Residuals Distribution

errors_home = y_train_home - rf_model_home.predict(X_train)
errors_away = y_train_away - rf_model_away.predict(X_train)

plt.figure(figsize=(10, 6))
sns.histplot(errors_home, color='blue', label='Home Score Errors', kde=True)
sns.histplot(errors_away, color='green', label='Away Score Errors', kde=True)
plt.xlabel('Residuals')
plt.ylabel('Frequency')
plt.title('Distribution of Residuals')
plt.legend()
plt.grid(True)
plt.show()

Predicted vs Training Data

import matplotlib.pyplot as plt

# Assuming you have actual scores in `nfl_train` for training predictions
plt.figure(figsize=(10, 6))
plt.scatter(nfl_train['score_home'], rf_model_home.predict(X_train), alpha=0.5, color='blue', label='Home Scores')
plt.scatter(nfl_train['score_away'], rf_model_away.predict(X_train), alpha=0.5, color='green', label='Away Scores')
plt.plot([nfl_train['score_home'].min(), nfl_train['score_home'].max()], [nfl_train['score_home'].min(), nfl_train['score_home'].max()], 'r--', label='Ideal Fit')
plt.xlabel('Actual Scores')
plt.ylabel('Predicted Scores')
plt.title('Predicted vs Actual Scores From Training Data')
plt.legend()
plt.grid(True)
plt.show()

Distribution of Scores

import seaborn as sns

plt.figure(figsize=(10, 6))
sns.histplot(nfl_train['score_home'], color='blue', label='Actual Home Scores', kde=True)
sns.histplot(rf_model_home.predict(X_train), color='cyan', label='Predicted Home Scores', kde=True)
plt.xlabel('Scores')
plt.ylabel('Frequency')
plt.title('Distribution of Actual vs Predicted Home Scores')
plt.legend()
plt.grid(True)
plt.show()

nfl_predict

	team_home	team_conference	team_division	team_conference_pre2002	team_division_pre2002	schedule_week	team_away	team_favorite_id	stadium	weather_detail	...	stadium_close	stadium_latitude	stadium_longitude	stadium_azimuthangle	stadium_elevation	schedule_season	schedule_playoff	stadium_neutral	predicted_score_home	predicted_score_away
13801	Kansas City Chiefs	AFC	AFC West	AFC	AFC West	1	Baltimore Ravens	Unknown	GEHA Field at Arrowhead Stadium	Unknown	...	2002.689877	39.048889	-94.483889	316.300000	264.900000	2024	0	0	17.57	20.50
13802	Tampa Bay Buccaneers	NFC	NFC South	NFC	NFC Central	1	Washington Commanders	Unknown	Raymond James Stadium	Unknown	...	2016.000000	27.975833	-82.503333	0.000000	5.800000	2024	0	0	19.12	18.85
13803	Cincinnati Bengals	AFC	AFC North	AFC	AFC Central	1	New England Patriots	Unknown	Paycor Stadium	Unknown	...	2002.689877	39.100000	-84.516670	192.485631	193.900000	2024	0	0	19.31	19.86
13804	Buffalo Bills	AFC	AFC East	AFC	AFC East	1	Arizona Cardinals	Unknown	Highmark Stadium	Unknown	...	2002.689877	42.889000	-78.890100	192.485631	178.000000	2024	0	0	23.37	18.59
13805	Los Angeles Chargers	AFC	AFC West	AFC	AFC West	1	Las Vegas Raiders	Unknown	SoFi Stadium	indoor	...	2002.689877	33.953450	-118.339200	338.200000	187.972944	2024	0	0	21.11	16.22
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
14067	Indianapolis Colts	AFC	AFC South	AFC	AFC East	18	Jacksonville Jaguars	Unknown	Lucas Oil Stadium	indoor	...	2002.689877	39.760056	-86.162806	26.600000	227.100000	2024	0	0	23.76	19.86
14068	New York Jets	NFC	AFC East	NFC	AFC East	18	Miami Dolphins	Unknown	MetLife Stadium	Unknown	...	2002.689877	40.813528	-74.074361	345.500000	2.100000	2024	0	0	16.41	21.15
14069	Baltimore Ravens	AFC	AFC North	AFC	AFC Central	18	Cleveland Browns	Unknown	M&T Bank Stadium	Unknown	...	2002.689877	39.278056	-76.622778	289.500000	47.500000	2024	0	0	21.40	21.36
14070	Green Bay Packers	NFC	NFC North	NFC	NFC Central	18	Chicago Bears	Unknown	Lambeau Field	Unknown	...	2002.689877	44.501389	-88.062222	0.000000	209.400000	2024	0	0	19.96	18.44
14071	Tennessee Titans	AFC	AFC South	AFC	AFC Central	18	Houston Texans	Unknown	Nissan Stadium	Unknown	...	2016.000000	36.166389	-86.771389	334.600000	182.900000	2024	0	0	21.88	18.90

271 rows × 33 columns